Performance & cost
7,061 rows, by stacks then stars
853 of these skill files have been read, by 797 distinct authors, and what they tell an agent counted. What Performance & cost authors agree on, and what they forbid
Internal standard selection optimization
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v1/skills/internal-standard-selection-optimization Skill
16★ repoUse when after batch correction of metabolomics QC samples using pooled study quality control (SQC) samples, when you have multiple candidate internal standards and need to determine which one stabilizes the compound/internal standard ratio for each compound.
Ion image embedding optimization
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v1/skills/ion-image-embedding-optimization Skill
16★ repoUse when you have 512-dimensional representation vectors output from ResNet18 encoders processing paired augmented ion images, and you need to prevent trivial solutions (representation collapse) during contrastive learning—specifically when optimizing for maximized similarity between augmentations.
Mass spectrometry benchmark design
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v1/skills/mass-spectrometry-benchmark-design Skill
16★ repoUse when when claiming that one mass spectrometry processing library achieves higher throughput than competitors, or when evaluating whether a new or optimized implementation delivers the expected computational efficiency gains.
Mass2motif parameter optimization
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v1/skills/mass2motif-parameter-optimization Skill
16★ repoUse when when you have a preprocessed bag-of-fragments corpus from tandem mass spectrometry spectra and need to train an MS2LDA model to discover Mass2Motifs.
Memory profiling and benchmarking
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v1/skills/memory-profiling-and-benchmarking Skill
16★ repoUse when when designing or optimizing backends that handle large MS datasets (mzML, mzXML, CDF files via MsBackendMzR), to verify that claimed memory advantages of on-disk or chunked approaches actually materialize in practice.
Metabolite detection and acquisition performance benchmarking
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v1/skills/metabolite-detection-and-acquisition-performance-benchmarking Skill
16★ repoUse when you have simulated or experimental mzML data from two or more fragmentation controllers (e.
Multi regime model performance comparison visualization
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v1/skills/multi-regime-model-performance-comparison-visualization Skill
16★ repoUse when you have paired microbiome-metabolome (or similar multivariate) datasets and want to quantify whether training on a superset of features (e.g., both annotated and unannotated metabolites) improves prediction accuracy for a well-defined target subset (e.
Multi window mass range parameter optimization
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v1/skills/multi-window-mass-range-parameter-optimization Skill
16★ repoUse when you observe sawtooth or discontinuous peak profiles in EICs after running tardisPeaks() on LC-MS data acquired with multiple overlapping or sequential m/z scan windows (e.g., positive and negative polarity scans, or data-independent acquisition with staggered windows).
Neural network threshold optimization
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v1/skills/neural-network-threshold-optimization Skill
16★ repoUse when after training a NeatMS neural network model on labelled peak data (High_quality, Low_quality, Noise) and you need to determine the optimal probability threshold for classifying peaks in your untargeted LCMS dataset.
Peak intensity threshold optimization
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v1/skills/peak-intensity-threshold-optimization Skill
16★ repoUse when you have loaded raw mass spectrometry spectral data (in MGF, MSP, mzML, or mzXML format) and need to decide which intensity threshold(s) to use for filtering out noise and low-abundance peaks.
Performance degradation quantification and analysis
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v1/skills/performance-degradation-quantification-and-analysis Skill
16★ repoUse when you have a pretrained model with documented performance on a bounded input domain (e.g., molecules ≤19 heavy atoms, sequences <1000 bp) and you need to establish whether and how much accuracy drops on held-out test cases outside that domain boundary.
Performance metric comparison across backends
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v1/skills/performance-metric-comparison-across-backends Skill
16★ repoUse when when you have access to a set of gallery or benchmark scripts executed across multiple plotting backends and need to quantify which backend delivers the fastest median execution time for specific mass spectrometry plot types (chromatogram, mobilogram, peakmap, peakmap-marginals, spectrum.
Precision recall optimization in spectral annotation
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v1/skills/precision-recall-optimization-in-spectral-annotation Skill
16★ repoUse when you have extracted fragmentation patterns from a collection of MS/MS spectra (using mineMS2) and have partitioned spectra into components via GNPS molecular networking (e.g., connected components, cliques, or high-similarity pairs with cosine > threshold).
Raw data throughput benchmarking
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v1/skills/raw-data-throughput-benchmarking Skill
16★ repoUse when when you have a raw mass spectrometry file (e.g., Thermo Orbitrap .raw) and need to establish the measured throughput of a spectral reading function (e.
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v1/skills/runtime-performance-profiling Skill
16★ repoUse when when you need to empirically validate that one mass spectrometry data processing library achieves higher throughput than competing alternatives. Specifically: you have multiple candidate libraries (e.
S4 class object memory profiling
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v1/skills/s4-class-object-memory-profiling Skill
16★ repoUse when when designing or optimizing S4-based data backends (such as MsBackend subclasses) and you need to decide whether to pre-populate all slots with complete data structures or defer initialization until data access.
Scalability extrapolation and throughput estimation
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v1/skills/scalability-extrapolation-and-throughput-estimation Skill
16★ repoUse when you have a new or modified LC-MS data processing tool and need to determine whether it can handle production-scale sample cohorts (50–100+ samples) on modest hardware (single-core CPU, ≤16 GB RAM).
Search mode configuration optimization
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v1/skills/search-mode-configuration-optimization Skill
16★ repoUse when when you have a calibrated FT-ICR mass spectrum (e.g., ESI-NEG mode) and need to decide between rapid single-assignment (first_hit=True) and exhaustive multi-assignment (first_hit=False) modes.
Silhouette analysis threshold optimization
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v1/skills/silhouette-analysis-threshold-optimization Skill
16★ repoUse when when you have a pre-computed hierarchical dendrogram from structural clustering (e.g., of LC-MS features based on m/z and retention time) and want to compare or validate the cluster assignments produced by a fixed constant-threshold method.
Software performance characterization and scaling analysis
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v1/skills/software-performance-characterization-and-scaling-analysis Skill
16★ repoUse when when a tool claims to be 'scalable' or 'performance-conscious' but lacks published performance benchmarks, or when you need to confirm that runtime and memory scale linearly (or predictably) with sample count before deploying the tool on large LC-MS datasets (e.g., >100 samples).
Tanimoto score threshold optimization
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v1/skills/tanimoto-score-threshold-optimization Skill
16★ repoUse when when you have a set of MS/MS spectra with ground-truth structural similarity labels (Tanimoto scores computed from molecular fingerprints) and need to choose a decision threshold for classifying spectrum pairs as 'chemically related' or 'unrelated'.
Targeted metabolite detection parameter optimization
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v1/skills/targeted-metabolite-detection-parameter-optimization Skill
16★ repoUse when when you have centroided .mzML LC–MS runs and a target list (compound ID, theoretical m/z, expected RT, polarity) but are uncertain whether your m/z and RT windows are wide enough to capture all targets without false positives.
Visualization comparative performance metrics
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v1/skills/visualization-comparative-performance-metrics Skill
16★ repoUse when you have execution-time data for the same set of plotting operations (e.g., chromatogram, spectrum, peakmap rendering) across two or more backend implementations (e.
Algorithm performance benchmarking
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v2/skills/algorithm-performance-benchmarking Skill
16★ repoUse when you have refactored or reimplemented a core computational method (e.g., entropy similarity calculation) and need to verify that the new implementation produces mathematically equivalent results to the original before deploying it to production.
Annotation benchmark performance evaluation
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v2/skills/annotation-benchmark-performance-evaluation Skill
16★ repoUse when after running an end-to-end annotation workflow (matching, clustering, filtering, and prioritization) on untargeted LC-MS peak tables, when you have access to a curated reference dataset (df.Ref) containing validated peak assignments for the same biological sample.
Api response latency measurement
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v2/skills/api-response-latency-measurement Skill
16★ repoUse when when annotating .msp files with metadata from multiple external web services and you need to monitor which services are slow or unreliable.
Backend performance benchmarking
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v2/skills/backend-performance-benchmarking Skill
16★ repoUse when when you have execution time data for visualization scripts across multiple backends (matplotlib, Bokeh, Plotly) and need to determine which backend offers the fastest median performance for specific mass spectrometry plot types (chromatogram, mobilogram, peakmap, peakmap-marginals.
Backend performance comparison
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v2/skills/backend-performance-comparison Skill
16★ repoUse when you have execution-time metrics (from a benchmark table or profiling logs) across multiple visualization backends for the same set of plots (e.
Backend performance tradeoff analysis
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v2/skills/backend-performance-tradeoff-analysis Skill
16★ repoUse when you are designing or optimizing an MsBackend implementation and need to decide whether to pre-populate the @spectraVars slot with all core spectra variable columns (mz, intensity, rtime, scanIndex, precursorMz, precursorIntensity, acquisitionNum, msLevel) at initialization time, or defer.
Baseline comparison performance benchmarking
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v2/skills/baseline-comparison-performance-benchmarking Skill
16★ repoUse when when you have implemented a novel annotation algorithm or network and must verify it outperforms a simpler reference approach on held-out or independent spectral data. Specifically: (1) you have ground-truth annotations for a common test set;
Batch construction parameter optimization
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v2/skills/batch-construction-parameter-optimization Skill
16★ repoUse when preparing labeled LC-MS peak data for neural network training and you need to decide whether class imbalance in your dataset should be preserved or corrected in batch construction. Use it particularly when your annotated peak dataset has unequal class distributions (e.
Batch script performance profiling
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v2/skills/batch-script-performance-profiling Skill
16★ repoUse when you have a collection of N scripts (e.g., 19 gallery examples) that must run on multiple backends or configurations, and you need to produce a reproducible benchmark table showing per-script execution times and aggregate totals.
Bayesian optimization acquisition function selection
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v2/skills/bayesian-optimization-acquisition-function-selection Skill
16★ repoUse when after fitting a Gaussian Process regression model to prior LC-MS gradient evaluations (where gradients are encoded as input and separation efficiency is output), use this skill to decide which candidate gradient to test next.
Beam search decoding optimization
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v2/skills/beam-search-decoding-optimization Skill
16★ repoUse when using Casanovo for de novo peptide sequencing on high-stakes datasets (immunopeptidomics, paleoproteomics, or monoclonal antibody discovery) where missing the correct sequence in the top-1 prediction is costly.
Benchmarking and comparative performance evaluation
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v2/skills/benchmarking-and-comparative-performance-evaluation Skill
16★ repoUse when you have trained a new machine learning model for chemical formula or adduct assignment from MS/MS spectra and need to assess whether it offers genuine performance gains over established baselines. Use it specifically when you have access to ground-truth annotations (e.
Binary cross entropy loss optimization
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v2/skills/binary-cross-entropy-loss-optimization Skill
16★ repoUse when when you have a pretrained spectrum encoder (TCN) and need to train a formula rescoring module that ranks candidate molecular formulas against MS/MS spectra.
Bipartite graph layout optimization
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v2/skills/bipartite-graph-layout-optimization Skill
16★ repoUse when after constructing a bipartite network graph with metabolites and proteins as nodes weighted by association strength and disease class, when you need to render the network visually for interpretation and publication.
Blind search mode parameter optimization
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v2/skills/blind-search-mode-parameter-optimization Skill
16★ repoUse when when you have tandem mass spectra from ribosomally synthesized peptides (RiPPs) and suspect the presence of unknown or non-standard post-translational modifications that would be missed by standard database search modes constrained to known modification classes.
Block layout analysis for io optimization
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v2/skills/block-layout-analysis-for-io-optimization Skill
16★ repoUse when when preparing NMR datasets for processing in NMRFx and the Dataset.createDataFile() method must choose among competing storage backends.
Cell type protein profiling integration
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v2/skills/cell-type-protein-profiling-integration Skill
16★ repoUse when you have co-registered IMC (protein imaging mass cytometry) and SIMS (secondary ion mass spectrometry for metabolites) data from the same tissue regions, cell segmentation masks, and need to assign cell types based on protein expression patterns, then overlay those assignments onto.
Chemical coverage optimization
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v2/skills/chemical-coverage-optimization Skill
16★ repoUse when your LC-MS feature table is incomplete or has low chemical coverage because traditional peak extraction algorithms (e.g., standard XCMS workflows) systematically miss features at m/z and retention time positions corresponding to known suspect compounds in your database.
Chunk wise data processing optimization
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v2/skills/chunk-wise-data-processing-optimization Skill
16★ repoUse when you have a large mass spectrometry dataset stored across multiple mzML, mzXML, or CDF files and need to perform operations (e.g., normalization, filtering, feature extraction) on the full dataset but memory constraints prevent loading all peak data simultaneously.
Classification model performance evaluation
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v2/skills/classification-model-performance-evaluation Skill
16★ repoUse when after fitting and optimizing a MB-PLS model on training data, apply this skill to the held-out test set (typically 10% of the original sample) to obtain unbiased performance estimates.
Classification performance evaluation
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v2/skills/classification-performance-evaluation Skill
16★ repoUse when after running inference on test mass spectrometry spectra with a trained deep learning model (e.g., PS2MS) to verify that class label predictions and confidence scores match expected reference outputs.
Classification performance metric evaluation
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v2/skills/classification-performance-metric-evaluation Skill
16★ repoUse when after training multiple classifiers (traditional ML and deep learning) on a preprocessed metabolomics feature matrix with labeled samples, and you need to quantify and compare their predictive performance to select the model or ensemble for downstream analysis.
Classification performance visualization
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v2/skills/classification-performance-visualization Skill
16★ repoUse when you have a CSV file with predicted probabilities and true binary labels from a classification model, and you need to evaluate classification performance across decision thresholds and communicate it via a standard diagnostic plot suitable for publication or presentation.
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v2/skills/cohort-performance-reporting Skill
16★ repoUse when you have NMR metabolite measurements from peripheral blood samples (plasma/serum) paired with processing delay metadata (pre-centrifugation and post-centrifugation times) and need to benchmark metabolic parameter stability across delay windows.
Cohort stratified metabolic performance analysis
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v2/skills/cohort-stratified-metabolic-performance-analysis Skill
16★ repoUse when when you have uploaded a pre-analytical data table containing sample metadata, processing delay annotations (pre- and post-centrifugation times), and paired NMR metabolomic measurements for a plasma or serum cohort, and you need to determine how processing delays impact metabolite.
Collision energy optimization for fragmentation
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v2/skills/collision-energy-optimization-for-fragmentation Skill
16★ repoUse when when you have N-Me derivatized unsaturated sterol lipid structures (as SMILES or molecular formula) and need to predict MS/MS fragmentation patterns with collision-energy-dependent m/z values and intensities for downstream CCS prediction or LC-IM-MS/MS library matching.
Comparative classifier performance assessment
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v2/skills/comparative-classifier-performance-assessment Skill
16★ repoUse when you have a labeled peak quality dataset (development set with ground-truth pass/fail labels), a defined set of peak-quality metrics (e.
Comparative performance evaluation across methods
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v2/skills/comparative-performance-evaluation-across-methods Skill
16★ repoUse when when you have developed a new spectral similarity scoring method and need to quantify its retrieval performance relative to established methods (e.g., modified cosine, Spec2Vec). Use this skill if you have: (1) a test set of spectra with ground-truth structural similarity labels (e.
Comparative performance profiling
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v2/skills/comparative-performance-profiling Skill
16★ repoUse when when you have implemented a new or optimized mass spectrometry data processing library and need to demonstrate its computational advantage over established alternatives (e.g., pymzML, pyOpenMS) on real proteomics data. Trigger on availability of: (1) a common input dataset (e.
Computational efficiency single pass vs repeated detection
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v2/skills/computational-efficiency-single-pass-vs-repeated-detection Skill
16★ repoUse when when processing aligned LC-MS data across multiple samples where the computational bottleneck is repeated peak-detection algorithm calls (one per sample per m/z value). Typical scenario: >10 samples with >1000 m/z values each, where N individual find_peaks invocations dominate runtime.
Computational performance benchmarking and timing measurement
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v2/skills/computational-performance-benchmarking-and-timing-measurement Skill
16★ repoUse when when you need to validate that a newly published bioinformatics pipeline meets stated performance claims, or when you must characterize how execution time scales with dataset size (sample count, peak count, or formula complexity) and optional analysis features (e.
Computational performance benchmarking
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v2/skills/computational-performance-benchmarking Skill
16★ repoUse when you have multiple candidate normalization methods for metabolomics data and need to choose one based on both statistical robustness and computational cost.
Computational resource profiling and benchmarking
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v2/skills/computational-resource-profiling-and-benchmarking Skill
16★ repoUse when when evaluating a new or updated version of a data processing tool (especially asari or similar LC-MS workflows) before production deployment, or when verifying claims about scalability, memory efficiency, or throughput on specific hardware classes (e.g., ≤16 GB RAM single-core systems).
Computational throughput measurement
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v2/skills/computational-throughput-measurement Skill
16★ repoUse when when you need to compare the computational efficiency of different mass spectrometry libraries on identical data and processing pipelines, or when you want to establish baseline throughput for a library version and validate claims of performance improvement.
Consensus clustering optimization
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v2/skills/consensus-clustering-optimization Skill
16★ repoUse when when you have a feature attribution matrix (e.g., microbe-metabolite interaction scores) from ensemble neural network training and need to group rows and columns into functionally coherent modules without pre-specifying cluster numbers.
Correlation threshold optimization
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v2/skills/correlation-threshold-optimization Skill
16★ repoUse when after computing pairwise correlations across all features in a dataset (especially those exceeding 10,000 features), before constructing the final network object.
Cosine annealing schedule optimization
HolobiomicsLab/asb-skill-collections/collections/metabolomics/v2/skills/cosine-annealing-schedule-optimization Skill
16★ repoUse when when training a regularized deep neural network for molecular property regression (e.g., retention time prediction on the METLIN SMRT dataset with 80,038+ samples), use cosine annealing warm restarts to escape plateaus and improve convergence.